Exposing properly with Dual ISO cameras

[Shirozina]

18 minutes ago, mercer said:

How’s the rolloff seem on the scopes at 100 iso? 

There is no 'rolloff' when exposing for digital using ETTR.

[mercer]

4 minutes ago, Shirozina said:

There is no 'rolloff' when exposing for digital using ETTR.

Okay, sorry... let me rephrase... how much information was there in the highlights as you ETTR’d... did you notice if a channel was clipping sooner than another channel?

[Shirozina]

Using the scopes after a WB there is nothing to recover after ETTR. Using the onboard zebras set at 100% it’s the same but the red channel clips first so it’s best to set the zebras at 95% to give some headroom if the highlights are not WB’d. Neither the zebras on the Ninja V or the onboard ones are accurate IME (for various reasons) so unless you have a monitor with RGB parade scopes I’d suggest take care with ETTR via zebras and if in doubt under expose or set them at 95 or even 90% Seems though that once ETTR is set you could ‘ride’ the ISO 400-1000 to get the look in terms of a light or dark  scene which is interesting but I need to test the noise to see if pushing 400ISO in post is the same as 1000 ISO in camera with the same scene and exposure.

 

[drm]

On 1/21/2019 at 6:44 AM, thephoenix said:

the thing is, if you follow his advice, that you would shoot 1200iso in daylight and 400 at night.

so why constructors spend time and money building dual iso cameras ? as most of us need high iso to shoot at night and low to shoot in daylight.

people doing photography are used to zone system, at least the one that shot print film.

you just need to set the exposure for the result you want more than for standards exposures.

with film we used to expose as wanted and we processed the films with order in order to complete the process

The exposure changes that he discussed in the video are clearly visible in the BMPCC4K footage, *if* you are not shooting RAW. It is the case that a lower ISO in the 100-1000 range and a lower ISO in the 1250-6400 range shows more details in the shadows. The ISOs in the higher end of each range do a better job preserving detail in the highlight areas. There is also a very dramatic change in the exposure from 1000 to 1250 as the camera changes from the low gain to the high gain circuit. ISO 1250 is substantially less noisy than 1000. I will have to compare again, but 1250 might even be less noisy than 800, but 800 would preserve bright details much better than 1250.

I guess that my point is that you do need to learn the subtle differences in this camera with regard to proper ISO selection for your scene, if you are not shooting RAW. I have not noticed the dramatic difference on my GH5 or GH5s cameras, despite their being dual native ISO cameras. It is fun to get to explore these

[kye]

4 hours ago, drm said:

The exposure changes that he discussed in the video are clearly visible in the BMPCC4K footage, *if* you are not shooting RAW. It is the case that a lower ISO in the 100-1000 range and a lower ISO in the 1250-6400 range shows more details in the shadows. The ISOs in the higher end of each range do a better job preserving detail in the highlight areas. There is also a very dramatic change in the exposure from 1000 to 1250 as the camera changes from the low gain to the high gain circuit. ISO 1250 is substantially less noisy than 1000. I will have to compare again, but 1250 might even be less noisy than 800, but 800 would preserve bright details much better than 1250.

I guess that my point is that you do need to learn the subtle differences in this camera with regard to proper ISO selection for your scene, if you are not shooting RAW. I have not noticed the dramatic difference on my GH5 or GH5s cameras, despite their being dual native ISO cameras. It is fun to get to explore these

Just a small thing but the GH5 isn't dual ISO, only the GH5S has that.

I wish my GH5 did have it, but sadly, no

[Shirozina]

12 hours ago, drm said:

It is the case that a lower ISO in the 100-1000 range and a lower ISO in the 1250-6400 range shows more details in the shadows. The ISOs in the higher end of each range do a better job preserving detail in the highlight areas.

This is not true. The low ISO's compress the shadows and put more tonal data in the highlights and the higher ISO's expand the shadows and compress the highlights. This is clearly evident on the scopes of a monitor or in an NLE. 

[drm]

12 hours ago, Shirozina said:

This is not true. The low ISO's compress the shadows and put more tonal data in the highlights and the higher ISO's expand the shadows and compress the highlights. This is clearly evident on the scopes of a monitor or in an NLE. 

The attached chart is directly from Blackmagic Design for the BMPCC4K. I may have worded my original message poorly, but Blackmagic shifts the middle grey point up or down based on the ISO when not recording RAW. With lower ISOs in the range, there are more points below the middle grey, with higher ISOs there are more points above the middle grey point.

[drm]

23 hours ago, kye said:

Just a small thing but the GH5 isn't dual ISO, only the GH5S has that.

I wish my GH5 did have it, but sadly, no

LOL...*that* is why my GH5 sucks in low light compared to the GH5s   I knew that only the GH5s had dual ISO, but it certainly didn't occur to me when I was writing that comment. There are a couple of other differences that sometimes trip me up, like timecode differences between the GH5 & GH5s. I have moved my GH5s cameras to A & B, and my GH5 is usually pulling duty as the C cam and as a cam for taking photos. For the last couple of shoots, I have even brought in one of the BMPCC4Ks and replaced the GH5, so the GH5 is only pulling duty for photos.

OH and I solved the BMPCC4K battery problems. I have a 158Wh v-mount battery on 15mm rods behind the camera. It runs pretty much all day now

[Shirozina]

9 hours ago, drm said:

The attached chart is directly from Blackmagic Design for the BMPCC4K. I may have worded my original message poorly, but Blackmagic shifts the middle grey point up or down based on the ISO when not recording RAW. With lower ISOs in the range, there are more points below the middle grey, with higher ISOs there are more points above the middle grey point.

Notice how the DR stays the same throughout the ISO ranges? This means BM is not varying the ISO like other cameras by digitally amplifying the whole signal but is altering the mid point in the response curve using a gamma like change. Cycle through the ISO at a constant exposure and see how the data moves on some scopes and you will understand. The above chart looks like they have made an exposure change for each ISO which is misleading. Look at any other camera DR vs ISO graph and the DR ramps down as ISO goes up just as in practice if you ETTR you need to change your exposure with ISO. Not so with the BM where the clipping point remains at the same exposure value throughout the ISO range. 

[drm]

2 hours ago, Shirozina said:

Notice how the DR stays the same throughout the ISO ranges? This means BM is not varying the ISO like other cameras by digitally amplifying the whole signal but is altering the mid point in the response curve using a gamma like change. Cycle through the ISO at a constant exposure and see how the data moves on some scopes and you will understand. The above chart looks like they have made an exposure change for each ISO which is misleading. Look at any other camera DR vs ISO graph and the DR ramps down as ISO goes up just as in practice if you ETTR you need to change your exposure with ISO. Not so with the BM where the clipping point remains at the same exposure value throughout the ISO range. 

@Shirozina thanks for your reply. It is a shame that we couldn't discuss this over a beverage It would be so much easier and faster.

I know that BM says that the dynamic range doesn't change as you change ISO within each range (100-1000, 1250-6400), they claim they are moving the middle grey point. That is why I posted the chart. 

My original comment was:

"...a lower ISO in the 100-1000 range and a lower ISO in the 1250-6400 range shows more details in the shadows. The ISOs in the higher end of each range do a better job preserving detail in the highlight areas."

to which you said:

"This is not true. The low ISO's compress the shadows and put more tonal data in the highlights and the higher ISO's expand the shadows and compress the highlights. This is clearly evident on the scopes of a monitor or in an NLE. " (emphasis added)

I posted the chart because you said that my original comment was not true.

QUESTION:

So, suppose that you are recording a landscape shot that includes clouds on a bright day and you want to emphasize detail in the clouds. What ISO should we use? With many other cameras, I would pick a low ISO to get the most dynamic range, then adjust the camera settings so the highlights are just below clipping. With the BMPCC4K, what ISO should we choose?

If we are recording RAW, I suggest that we choose the 100-1000 ISO range to maximize dynamic range then expose so we aren't clipping the highlights.

What if we are not recording RAW? What ISO would you suggest that someone use when recording this scene on the BMPCC4K, when not recording in RAW, in order to have the most detail in the highlights?

 

[Shirozina]

Hum - beer......?

If you need to retain highlights without clipping then first of all ETTR is naturally the best exposure method and if in addition to this you want a lot of tonal information in those highlight areas then a low ISO with it's flatter response curve allocates more data in the upper end of the response curve. DR remains the same so it's just about placing the most data where you want the most tonal detail and with the P4k once you have got an ETTR exposure just under clipping you can change the ISO and highlights will remain unclipped (as long as you use scopes and not the Zebras). On the other hand if you want to avoid clipping the clouds but want to show a lot of shadow detail then I'd be setting a high ISO that allocates more data in the lower levels and compresses the highlights into less data. With RAW it doesn't matter as long as you ETTR at native ISO but RAW is not truly RAW on the P4k as it still has a tonal response curve baked into the data so I'd still want to get this curve as close as possible in camera to the required output. Obviously for maximum DR I'd use the lower of the 2 native ISO ranges. I need to photograph a stepped density chart from black to white using the same exposure but different ISO to show the change in the response curves. Again I have to state that to do this properly you need to use scopes as ETTR with the Zebras away from 400 and 3200 doesn't appear to be very accurate. 

[kye]

8 hours ago, Shirozina said:

I need to photograph a stepped density chart from black to white using the same exposure but different ISO to show the change in the response curves.

Isn't that what John does in the original video?

[Shirozina]

9 hours ago, kye said:

Isn't that what John does in the original video?

Yes at around 5'36" he shows the effect of increasing ISO but says 'everything steps up' when he increases ISO but if you follow the scopes it's only the mid and shadows that increase and the highlights stay the same. Notice the highlights do move BUT the max IRE available at each ISO is not fixed and starts at about 80% at 100 iso (which will be normalised to 100IRE during grading) and increases up to 100% at 1000 ISO . His other tests are misleading as he's changing exposure to compensate for ISO changes which is not required as the digital ISO changes are a log curve adjustment where black and white points are fixed. This is certainly a different approach to ISO and shows the problem with the concept of ISO when applied to a LOG capture. ISO was and is a legacy from the days of film when the response curve of film was not log and the ISO was rated at a midpoint ( mid grey) along the middle flat part of the graph. With LOG the mid point is elevated and bears no relation to a mid point ISO hence why Native ISO for camera shooting LOG is always much higher than standard picture profiles.

[kye]

44 minutes ago, Shirozina said:

Yes at around 5'36" he shows the effect of increasing ISO but says 'everything steps up' when he increases ISO but if you follow the scopes it's only the mid and shadows that increase and the highlights stay the same. Notice the highlights do move BUT the max IRE available at each ISO is not fixed and starts at about 80% at 100 iso (which will be normalised to 100IRE during grading) and increases up to 100% at 1000 ISO . His other tests are misleading as he's changing exposure to compensate for ISO changes which is not required as the digital ISO changes are a log curve adjustment where black and white points are fixed. This is certainly a different approach to ISO and shows the problem with the concept of ISO when applied to a LOG capture. ISO was and is a legacy from the days of film when the response curve of film was not log and the ISO was rated at a midpoint ( mid grey) along the middle flat part of the graph. With LOG the mid point is elevated and bears no relation to a mid point ISO hence why Native ISO for camera shooting LOG is always much higher than standard picture profiles.

This sure is a pretty complicated topic.

I get that if you shoot in Prores it is applying a log curve of some kind which shifts the middle point more than the highlights and shadows.  A variant of BMD Film perhaps.

But i'm still confused about RAW - is the RAW signal still in Linear?  Sensors see in linear, so if they're somehow changing it then surely that's no longer RAW?

[Shirozina]

12 minutes ago, kye said:

This sure is a pretty complicated topic.

I get that if you shoot in Prores it is applying a log curve of some kind which shifts the middle point more than the highlights and shadows.  A variant of BMD Film perhaps.

But i'm still confused about RAW - is the RAW signal still in Linear?  Sensors see in linear, so if they're somehow changing it then surely that's no longer RAW?

BM RAW is not like other cameras RAW as it has a curve built into the data so it's not 'true' RAW in the same way that a still RAW from a Canon, Nikon, Sony etc is.